A root appearing Notes on Names (NN) from 1957 serving as an explanation for the element S. thorn “steadfast” in the name S. Arathorn, first appearing in a rejected page with variants √STOR and √THOR (PE17/113-114) and then later as only √THOR (PE17/113). This root may be connected to Q. torna “hard” in notes on Quenya intensive forms written between the first and second edition of The Lord of the Rings, where it was an element in Q. tornanga “hard iron”, with intensive forms aristorna, anastorna that imply derivation from √STOR (PE17/56).
Neo-Eldarin: For purposes of Neo-Eldarin, I would assume this root is √STOR to avoid conflict with ᴹ√THOR(ON), the basis for “eagle” words.
The root √BER appeared in later writing as the basis for marriage words, and Quenya marriage words began with ver- for much of Tolkien’s life, dating all the way back in the Qenya Lexicon of the 1910s. However, there are many ways to produce a medial r in Quenya, and Tolkien experimented with more or less all of them in the formation of this root, as seen by the shifting forms in other branches in the Elvish languages.
The earliest form of this root appeared as √VEŘE in the Qenya Lexicon of the 1910s (QL/101), where the Ř indicated the actual primitive form was VEÐE (or ɃEÐE), as confirmed by its Gnomish equivalent Bedh- (QL/101; GL/22). This early root was unglossed, but its derivatives all had to do with marriage, including ᴱQ. veru/G. †benn/G. bedhron “husband”, ᴱQ. †veri/ᴱQ. vesse/G. bess “wife” and ᴱQ. vesta-/G. benna- “to wed”.
There are indications this root may have initially been ᴹ√BED in The Etymologies of the 1930s, based on primitive forms like ᴹ✶bedū “spouses” (Ety/LEP). But the main “marriage” entry in The Etymologies was ᴹ√BES “wed”, with derivatives like ᴹQ. vesta- “to wed”, ᴹQ. venno/N. benn “husband” and ᴹQ. vesse/N. bess “wife”. One interesting feature about this conceptual stage is that the Noldorin words for “husband” and “wife” drifted in meaning to become simply “man” and “woman”, and new words N. herven and herves were coined for “husband” and “wife”.
In various etymological notes from the late 1960s, Tolkien introduced a new root √BER for marriage. His longest description of it was in a 1969 note:
> √BER “to mate, be mated, joined in marriage”: Q. verya (intr.) “to marry (of husband and wife), be joined to” (veryanen senna); verta (tr.) “to give in marriage (a) to (b), or to take as husband or wife (to oneself)”; verū > veru “husband”; verī > veri “wife” (VT49/45).
The forms Q. veru/veri “husband/wife” were restorations of the Early Qenya forms from the 1910s.
It is not entirely clear when Tolkien made this change to the root, but in The Road Goes Ever On published in 1967, he said that “[S.] bereth actually meant ‘spouse’, and is used of one who is ‘queen’ as spouse of a king” (RGEO/66), strongly hinting that this word was derived from √BER “wed” at that conceptual stage. Earlier on, N./S. bereth “queen” was derived from primitive ✶barathī, an etymology Tolkien was still using in notes from the late 1950s or early 1960s (PE17/22-23). As discussed in the entry on Elbereth, this etymology presented some phonological difficulties in Sindarin, which may have motivated Tolkien to introduce a new etymology for S. bereth, switching to the root √BER for marriage in the process. My best guess is that this switch happened sometime in the mid-1960s, but it could have been earlier.
Neo-Eldarin: For purposes of Neo-Eldarin writing, I think it is best to ignore the introduction of the root √BER “wed”, and keep using ᴹ√BES “wed” from The Etymologies. As a root for marriage, √BER presents a number of problems:
It clashes with ᴹ√BER “valiant” from the 1930s, the basis for the name S. Beren.
It invalidates a number of Noldorin words for husband and wife with no obvious replacements.
In Sindarin it is difficult to keep derivatives of √BER “wed” from colliding with other words like beria- “to protect”, bertha- “to dare” and beren “bold”, though technically these are all Noldorin words.
The main disadvantage of ignoring √BER “wed” is that it leaves us with no good etymology for the name S. Elbereth. It is quite possible to solve the above problems with √BER by coining new words, but I prefer to keep using the older words as they are quite popular in Neo-Sindarin. It would be relatively easy to coin a new etymology for Elbereth instead, perhaps from ✱elen-baratthī with the longer cluster inhibiting i-intrusion. See the entry on S. bereth for further discussion.